Contact us


    Collection and Use of Personal Information. *


    Deep Bio Presents its Novel Deep Learning-based Gleason Grading System in npj Digital Medicine

    July 1, 2021

    Deep Bio developed a novel deep learning-based prostate cancer detection model that demonstrates superb performance trained by a weakly supervised learning method 

    Seoul, South Korea, July 1, 2021 – Deep Bio today announced that it published research results on a study, ‘Yet Another Automated Gleason Grading System (YAAGGS),’ in npj Digital Medicine which is a part of the prestigious international science and technology journals series, Nature Partner Journals. YAAGGS is Deep Bio’s original deep learning-based Gleason grading system that detects prostate cancer and classifies cancer severity.

    The study presents a new weakly-supervised deep learning-based prostate cancer detection model. The study outlines training and validation of the model using whole slide images (WSIs) and their corresponding hospital diagnoses. To demonstrate the performance of YAAGS, a total of 7,600 H&E (Hematoxylin and Eosin)-stained prostate core needle biopsy samples images were used. The specimens were provided by Hanyang University Medical Center and Korea University Guro Hospital in South Korea.

    In regard to cancer detection performance, the YAAGGS model exhibited a ROC AUC value of 0.983. Its sensitivity and specificity were 93.6%, and 96.0%, respectively. For the grade group prediction, analysis showed a Cohen’s kappa score of 0.650 and a quadratic-weighted kappa score of 0.897. The study demonstrated that the model can detect and grade the severity of prostate cancer just as well as a board-certified pathologist.

    Detailed annotations on regions of interest specifying the severity of cancer are known to be essential for the development of deep learning-based cancer diagnostic algorithms. However, the manual annotation process is highly inefficient in terms of development time and cost. The novel approach to deep learning showcased in this paper can be used to expedite the development of additional diagnostic models.

    Sun Woo Kim, the CEO of Deep Bio, said, “The YAAGGS study suggests a new method of deep learning that can significantly reduce the time and cost required to develop a cancer diagnostic solution. At the same time, YAAGGS also demonstrated superb performance, solidifying Deep Bio’s position as a leader in AI cancer diagnostics.” He also added, “Deep Bio will continue to innovate and work to optimize the digital pathology workflow to improve patient care.”

    About Deep Bio

    Deep Bio Inc. is an AI biotech company with in-house expertise in deep learning, pathology, life sciences, and pharmacotherapeutics. As the country’s first to obtain KFDA approval of an AI-based cancer pathology solution, Deep Bio envisions a suite of AI-based IVD SaMDs (In Vitro Diagnostics Software as a Medical Device) for diagnosis and prognosis prediction of multiple cancers. Deep Bio is actively engaged in the research space and participating in ongoing collaborations with top US medical centers.

    DeepDx® Prostate is a clinically-validated AI for prostate core needle biopsy tissue image analysis. Whole-slide images (WSIs) of H&E-stained biopsy tissue specimens are analyzed for prostate cancer, Gleason scores and grade group. Extensively tested at a US CLIA lab (>240k cores in 2020), DeepDx® Prostate can alleviate the shortage of pathologists and the resultant increase in workload, while reducing diagnostic subjectivity and variability. To learn more, visit